#include <ctype.h>
#include <stdio.h>
-#define TAG_MAX_LEN 64
-
#define SPACECHR " \t\r\n\f"
-#define DEBUG(x) x
-#if HAVE_SYS_TYPES_H
-#include <sys/types.h>
-#endif
+namespace metaproxy_1 {
+ class HTMLParser::Rep {
+ friend class HTMLParser;
+ public:
+ void parse_str(HTMLParserEvent &event, const char *cp);
+ void tagText(HTMLParserEvent &event,
+ const char *text_start, const char *text_end);
+ int tagEnd(HTMLParserEvent &event,
+ const char *tag, int tag_len, const char *cp);
+ int tagStart(HTMLParserEvent &event,
+ int *tag_len, const char *cp, const char which);
+ int tagAttrs(HTMLParserEvent &event,
+ const char *name, int len,
+ const char *cp);
+ Rep();
+ ~Rep();
+ int m_verbose;
+ };
+}
namespace mp = metaproxy_1;
-mp::HTMLParser::HTMLParser()
+mp::HTMLParser::Rep::Rep()
{
+ m_verbose = 0;
}
-mp::HTMLParser::~HTMLParser()
+mp::HTMLParser::Rep::~Rep()
{
}
-static void parse_str(mp::HTMLParserEvent & event, const char * str);
+mp::HTMLParser::HTMLParser() : m_p(new Rep)
+{
+}
-void mp::HTMLParser::parse(mp::HTMLParserEvent & event, const char *str) const
+mp::HTMLParser::~HTMLParser()
{
- parse_str(event, str);
}
-//static C functions follow would probably make sense to wrap this in PIMPL?
+void mp::HTMLParser::set_verbose(int v)
+{
+ m_p->m_verbose = v;
+}
+
-static char* dupe (const char *buff, int len)
+void mp::HTMLParser::parse(mp::HTMLParserEvent & event, const char *str) const
{
- char *value = (char *) malloc (len + 1);
- assert (value);
- memcpy (value, buff, len);
- value[len] = '\0';
- return value;
+ m_p->parse_str(event, str);
}
-static int skipSpace (const char *cp)
+static int skipSpace(const char *cp)
{
int i = 0;
- while (cp[i] && strchr (SPACECHR, cp[i]))
+ while (cp[i] && strchr(SPACECHR, cp[i]))
i++;
return i;
}
-static int skipName (const char *cp, char *dst)
+static int skipName(const char *cp)
{
int i;
- int j = 0;
- for (i=0; cp[i] && !strchr (SPACECHR "/>=", cp[i]); i++)
- if (j < TAG_MAX_LEN-1)
- {
- dst[j] = tolower(cp[j]);
- j++;
- }
- dst[j] = '\0';
+ for (i = 0; cp[i] && !strchr(SPACECHR "/>=", cp[i]); i++)
+ ;
return i;
}
-static int skipAttribute (const char *cp, char *name, const char **value, int *val_len)
+static int skipAttribute(const char *cp, int *attr_len,
+ const char **value, int *val_len)
{
- int i = skipName (cp, name);
+ int i = skipName(cp);
+ *attr_len = i;
*value = NULL;
if (!i)
- return skipSpace (cp);
- i += skipSpace (cp + i);
+ return skipSpace(cp);
+ i += skipSpace(cp + i);
if (cp[i] == '=')
{
int v0, v1;
i++;
- i += skipSpace (cp + i);
+ i += skipSpace(cp + i);
if (cp[i] == '\"' || cp[i] == '\'')
{
char tr = cp[i];
v0 = ++i;
while (cp[i] != tr && cp[i])
- i++;
+ i++;
v1 = i;
if (cp[i])
i++;
else
{
v0 = i;
- while (cp[i] && !strchr (SPACECHR ">", cp[i]))
+ while (cp[i] && !strchr(SPACECHR ">", cp[i]))
i++;
v1 = i;
}
*value = cp + v0;
*val_len = v1 - v0;
}
- i += skipSpace (cp + i);
+ i += skipSpace(cp + i);
return i;
}
-static int tagAttrs (mp::HTMLParserEvent & event,
- const char *tagName,
- const char *cp)
+int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
+ const char *name, int len,
+ const char *cp)
{
- int i;
- char attr_name[TAG_MAX_LEN];
- const char *attr_value;
- int val_len;
- i = skipSpace (cp);
- while (cp[i] && cp[i] != '>')
+ int i = skipSpace(cp);
+ while (cp[i] && cp[i] != '>' && cp[i] != '/')
{
- int nor = skipAttribute (cp+i, attr_name, &attr_value, &val_len);
+ const char *attr_name = cp + i;
+ int attr_len;
+ const char *value;
+ int val_len;
+ int nor = skipAttribute(cp+i, &attr_len, &value, &val_len);
i += nor;
if (nor)
{
- DEBUG(printf ("------ attr %s=%s\n", attr_name, dupe(attr_value, val_len)));
- event.attribute(tagName, attr_name, attr_value, val_len);
+ if (m_verbose)
+ printf ("------ attr %.*s=%.*s\n", attr_len, attr_name,
+ val_len, value);
+ event.attribute(name, len, attr_name, attr_len, value, val_len);
}
else
{
- if (!nor)
- i++;
+ i++;
}
}
return i;
}
-static int tagStart (mp::HTMLParserEvent & event,
- char *tagName, const char *cp, const char which)
+int mp::HTMLParser::Rep::tagStart(HTMLParserEvent &event,
+ int *tag_len,
+ const char *cp, const char which)
{
- int i = 0;
- i = skipName (cp, tagName);
- switch (which)
+ int i;
+ switch (which)
{
- case '/' :
- DEBUG(printf ("------ tag close %s\n", tagName));
- event.closeTag(tagName);
- break;
- case '!' :
- DEBUG(printf ("------ dtd %s\n", tagName));
- break;
- case '?' :
- DEBUG(printf ("------ pi %s\n", tagName));
- break;
- default :
- DEBUG(printf ("------ tag open %s\n", tagName));
- event.openTagStart(tagName);
- break;
+ case '/':
+ i = skipName(cp);
+ *tag_len = i;
+ if (m_verbose)
+ printf("------ tag close %.*s\n", i, cp);
+ event.closeTag(cp, i);
+ break;
+ case '!':
+ for (i = 0; cp[i] && cp[i] != '>'; i++)
+ ;
+ *tag_len = i;
+ event.openTagStart(cp, i);
+ if (m_verbose)
+ printf("------ dtd %.*s\n", i, cp);
+ break;
+ case '?':
+ for (i = 0; cp[i] && cp[i] != '>'; i++)
+ ;
+ *tag_len = i;
+ event.openTagStart(cp, i);
+ if (m_verbose)
+ printf("------ pi %.*s\n", i, cp);
+ break;
+ default:
+ i = skipName(cp);
+ *tag_len = i;
+ if (m_verbose)
+ printf("------ tag open %.*s\n", i, cp);
+ event.openTagStart(cp, i);
+
+ i += tagAttrs(event, cp, i, cp + i);
+
+ break;
}
return i;
}
-static int tagEnd (mp::HTMLParserEvent & event, const char *tagName, const char *cp)
+int mp::HTMLParser::Rep::tagEnd(HTMLParserEvent &event,
+ const char *tag, int tag_len, const char *cp)
{
int i = 0;
+ int close_it = 0;
while (cp[i] && cp[i] != '>')
+ {
+ if (cp[i] == '/')
+ close_it = 1;
i++;
+ }
if (cp[i] == '>')
{
- event.anyTagEnd(tagName);
+ event.anyTagEnd(tag, tag_len, close_it);
i++;
}
return i;
}
-static void tagText (mp::HTMLParserEvent & event, const char *text_start, const char *text_end)
+void mp::HTMLParser::Rep::tagText(HTMLParserEvent &event,
+ const char *text_start, const char *text_end)
{
if (text_end - text_start) //got text to flush
{
- DEBUG(printf ("------ text %s\n", dupe(text_start, text_end-text_start)));
+ if (m_verbose)
+ printf("------ text %.*s\n",
+ (int) (text_end - text_start), text_start);
event.text(text_start, text_end-text_start);
}
}
-static void parse_str (mp::HTMLParserEvent & event, const char *cp)
+void mp::HTMLParser::Rep::parse_str(HTMLParserEvent &event, const char *cp)
{
const char *text_start = cp;
const char *text_end = cp;
if (cp[0] == '<' && cp[1]) //tag?
{
char which = cp[1];
- if (which == '/') cp++;
- if (!strchr (SPACECHR, cp[1])) //valid tag starts
+ if (which == '/')
+ cp++;
+ if (!strchr(SPACECHR, cp[1])) //valid tag starts
{
- tagText (event, text_start, text_end); //flush any text
- char tagName[TAG_MAX_LEN];
+ int i = 0;
+ int tag_len;
+
+ tagText(event, text_start, text_end); //flush any text
cp++;
- if (which == '/')
- {
- cp += tagStart (event, tagName, cp, which);
- }
- else if (which == '!' || which == '?') //pi or dtd
- {
- cp++;
- cp += tagStart (event, tagName, cp, which);
- }
- else
- {
- cp += tagStart (event, tagName, cp, which);
- cp += tagAttrs (event, tagName, cp);
- }
- cp += tagEnd (event, tagName, cp);
+ i += tagStart(event, &tag_len, cp, which);
+ i += tagEnd(event, cp, tag_len, cp + i);
+ cp += i;
text_start = cp;
text_end = cp;
continue;
cp++;
text_end = cp;
}
- tagText (event, text_start, text_end); //flush any text
+ tagText(event, text_start, text_end); //flush any text
}
/*
projects / metaproxy-moved-to-github.git / blobdiff